Introduction
The first of its kind, this term’s ISSS608 Visual Analytics and Applications course led by Professor Kam Tin Seong builds us towards the highly prestigious IEEE Vast Challenge 2022.
There are 4 challenges provided, targeting different aspects of life in the fictional town Engagement. Each team is to choose one challenge to create a visualisation solution that will be hosted on a web application using the R package - Shiny. We chose to analyse the Economics of Engagement, and created a dashboard that allows the user to visually analyse the interplay between employee (or residents) and business (or employers) before delving deeply into exploring either.
Brief Information About Competition
This year, the competition features the fictional town Engagement. We are given information from 2022-03-01T00:00:00Z to 2022-03-04T14:25:00Z about the 1010 participants who reside there.
Datasets
In general, there are 13 datasets.
| Dataset | Description | |
|---|---|---|
| 1 | Participant StatusLogs | A collection of 72 Comma-Separated Values (CSV) files, each containing information about all participants’ day-to-day activities. Each CSV file contains information about the participant’s current location, hunger status, sleep status, apartment, available balance, job, financial status, daily food budget and extra weekly budget. |
| 2 | Apartments | Information about the rental cost, max occupancy, number of rooms, and location of buildings in Engagement. |
| 3 | Buildings | Information about the location, types and maximum occupancy of buildings in Engagement. |
| 4 | Employers | Information about the location of employers in Engagement. |
| 5 | Jobs | Information about employers, hourly rate, start time, end time, days to work and education requirements that employees in Engagement have to meet. |
| 6 | Participants | Overarching information about each participant, including household size, whether they have kids, age, education levels, interest groups and joviality. |
| 7 | Pubs | Information about the hourly cost of operation, mac occupancy and location of pubs in Engagement. |
| 8 | Restaurants | Information about the cost of food, maximum occupancy and location of restaurants in Engagement. |
| 9 | Schools | Information about the monthly costs, maximum enrollment and location of schools in Engagement. |
| 10 | CheckinJournal | Information about the time participants check into different venues in Engagement. |
| 11 | Financial Journal | Information about the time and amount participants spend on certain types of items (ie Education, Food, Recreation, Rent Adjustment, Shelter and Wage) |
| 12 | SocialNetwork | Information about the interaction between 2 participants in Engagement. |
| 13 | TravelJournal | Information about the all travel start and end time and duration, their purpose, the check-in and check-out timing associated with such travel, and the amount a participant starts and ends off with during such travel. |
Challenges
There were 4 challenges provided. Each one asks us to choose a perspective to evaluate Engagement by.
They are:
Demographics and Relationships
Patterns of Life
Economic
Combining analysis from Challenges 1 to 3
Project Requirements
Challenge Chosen
We chose Challenge 3 (Economics), and its accompanying sub-questions.
| Overarching Question | |
|---|---|
| Over time, are businesses growing or shrinking? How are people changing jobs? Are standards of living improving or declining over time? | |
| Sub -Questions | |
| 1 | Over the period covered by the dataset, which businesses appear to be more prosperous? Which appear to be struggling? Describe your rationale for your answers. Limit your response to 10 images and 500 words. |
| 2 | How does the financial health of the residents change over the period covered by the dataset? How do wages compare to the overall cost of living in Engagement? Are there groups that appear to exhibit similar patterns? Describe your rationale for your answers. Limit your response to 10 images and 500 words. |
| 3 | Describe the health of the various employers within the city limits. What employment patterns do you observe? Do you notice any areas of particularly high or low turnover? Limit your response to 10 images and 500 words. |
Visualisation Solution
In order to visualise the changes systematically, we take a multi-prong approach. More specifically, there are 4 “prongs” — one for each question. All 4 prongs will be parked under a unifying dashboard, with each prong being accessible by a tab. The questions will be arranged sequentially, starting from Overarching Questions (ie Overview), Sub-Question 1 (ie Business), Sub-Question 2 (ie Residents) and Sub-Question 3 (ie Employee-Employer).
We arrange our tabs in this manner because we want to guide our user into a top-down understanding of the Economic in Engagement. The Overview tab provides the user with a general understanding of the economic situation in Engagement. Using the insight he or she gathered and the hypotheses created during, the user then explores the business and employee (or residents) side individually. With that new-found knowledge, he or she can check out the relationship between employee and employer (or business) to develop a full understanding of Engagement’s economy.
Overview
This will be the landing page for all users where they find out more about the Vast Challange 3 and user instructions to navigate through the shiny app application.
We will also include short profile of each member.
Sub-Question 1
Over the period covered by the dataset, which businesses appear to be more prosperous? Which appear to be struggling? Describe your rationale for your answers.
Upon clicking on the ‘Business’ tab, the following default view will be displayed.
The 3 features in which users can play around with are
| Feature | Function |
|---|---|
| Category (Radio button) | User has the option to view either restaurants or pubs. |
| Scenario (Dropdown) | User can select the type of scenario they would like to visualize. Select ‘Top 5’ or ‘Top 10’ to view the top 5 or top 10 businesses to see which companies are prospering. |
| Date (Range Slider) | User has the ability to scope to the target range date. |
When the user clicked on a particular bar chart, the corresponding line chart to its right will be highlighted. This action of clicking on the bar chart will also result in 2 more charts to appear, a boxplot and a heatmap.
Rationale
Total Sales 
The main objective of the first question in challenge 3 is to
identify which businesses are prospering or struggling. As there is no
sales and expense data found on the Restaurants.csv and
Pubs.csv, the only way to obtain sales volume is to compute
via the customer’s spending on these businesses. Prosperous businesses
will refer to companies that are making more sales whereas struggling
businesses will refer to companies that have lower sales. As the sales
volume fluctuates sporadically day-to-day, to determine the top 5
companies, the total overall sales are calculated and the 5 companies
with the most sales are selected.
Monthly Sales 
As the Total Sales Chart only compares at the overall sales volume, it provides no insights on how the businesses are performing overtime. Therefore, the Monthly Sales chart visualizes how the companies are performing at a monthly basis. We see that in general, there is a sharp decline in sales from March 2022 to April 2022. Businesses continues to dip till June 2022 after which the sales fluctuates.
Average Spending of Customers 
To give insights to users on the weekday sales of the selected businesses. Are customers spending more on the weekday or the weekends?
Occupancy heatmap 
The Occupancy Heatmap provides
vizualisation on the average occupancy rate for the selected company
across the weekdays at hourly intervals. Users are able to see which are
the peak and non-peak periods.
Sub-Question 2
Upon clicking on the ‘Resident’ tab, the following default view will be displayed.
Our visualisation solution for Sub-Question 2 is made up of 4 parts:
| Colour | Name | Function |
|---|---|---|
| Grey | Heat Map | Each row in the heat map represents a participant. Each column in the heat map represents a time period. To adjust the number of participants in the heat map, we can use the “Top X filter” in the Heat Map Controller. More importantly, the magnitude of the colour is representative of the amount of balance or debt. For example, a participant with more debt in a certain time period will have a denser red-coloured grid than a user with less debt. By hovering over the participant tab, the user can get more information about each user that goes beyond the scope of this question. |
| Blue | R epresentative Violin Plot and Box Plot | While the heat map is great, it does not encapsulate any statistical information. We gather such information for the total, average and customised (based on chosen customers) values of income and cost of living. To represent the distribution of both income and cost of living, we use a violin plot represented by 2 coloured rectangles — one for each. We also layer a boxplot over in order to catch the mea, variable, maximum and minimum values within each set up. |
| Green | Filter | Our controls provide the user with more flexibility when it comes to manipulating Engagement’s economic data. The user inputs the “Category” he or she wants, and then chooses a range of values wanted. He or she can also choose to look at particular periods of time by setting the start time and end time of analysis using the time scrub. |
| Orange | Heat Map Controller | Patterns are easier when we sort the participants according to their similarities. In order to determine the closest similarity between 2 participants, we use a leaf sort. On top of that, the user can also control the heatmap using the “Top X filter” and “Time filter” as mentioned in the “Representative Violin Plot and Box Plot” part of this paper. |
Let’s see how this visualisation solution can be used to solve the 3 pointers.
Pointer 1: “How does the financial health of the residents change over the period covered by the dataset?”
The heat map clearly demonstrates this change over time. If the user wants a snapshot value, he or she can use the violin plot and boxplot.
Pointer 2: “How do wages compare to the overall cost of living in Engagement?”
To check wages to overall cost, we can simply use the violin plots and box plots.
Pointer 3: “Are there groups that appear to exhibit similar patterns?”
The heat map demonstrates the individual changes in financial status and makes any patterns very obvious. With the ranking algorithm, we get to pull groups of participants that are similar in one way or another close to one another, making it easier for us to check if there is an issue.
Sub-Question 3
Describe the health of the various employers within the city limits. What employment patterns do you observe? Do you notice any areas of particularly high or low turnover?
Upon clicking on the ‘Employer-Employee’ tab, the following default view will be displayed.
There are 6 features within this tab:
| Feature | Function |
|---|---|
| Map/Chart | User has the option to choose between map or chart view. |
| Entity View | User see more information (employer ID, job ID) within each building ID. |
| City Map | User will be able to see turnover rate and number of employees for each employer in each building. |
| Control Panel | User filter through the options to different scenario. |
| Timeline | User has the ability to scope to the target range date, there will also be a function to auto play and see the changes over time. |
| Multiple Employers | User can select maximum 5 employers to see change in their number of employees over time. |
There are 4 charts for the users to explore, each will have tooltip when hover over:
| Feature | Function |
|---|---|
| Line Chart1 | Percentage % Change of Employees For Each Work Week |
| Line Chart2 | Number of Employees For Each Work Week |
| Boxplot | Wage Distribution for Different Education Level |
| Heatmap | Each cube = Employer, Size = No. of Employee, Color = Age of Employees |
The above aims to address:
Pointer 1: Describe the health of the various employers within the city limits.”
This refers to the general state of each employer which is described by the employment pattern and turnover rate from the chart and map view respectively.
Pointer 2: “What employment patterns do you observe?
The wage and age distribution under the chart view give hints on possible patterns.
Pointer 3: “Do you notice any areas of particularly high or low turnover?
The different color intensity of turnover rate gives user a sensing across all employers with accompanying line charts for more statistical information.
Proposed Packages
These are the R packages we are planning on using. This is subject to change as the requirements of the project from our end might change.
| Package Name | Use |
|---|---|
| shiny and shiny dashboard | To build an interactive web visualisation. |
| shinythemes | To make our shiny website more aesthetic. |
| tidyverse | A must-have for data manipulation and exploration. |
| plotly | To create all bar charts and graphs. |
| ggplot2 | To plot our visualisations and conduct Exploratory Data Analysis. |
| treemap | To display hierarchical data as a set of nested rectangles. |
| ggrepel | To repel overlapping text labels away from each other and away from the data points that they label. |
| lubridate, zoo | To make it easier to work with dates and times. |
| gapminder | For data wrangling and visualization |
| gganimate | To create animated ggplots. |
| ggiraph | To create dynamic ggplot graphs. |